Brain cholesterol in normal and pathological aging

Introduction

Cholesterol is essential for the function of most eukaryotic cells. As a major component of the plasma membrane, cholesterol plays a key role in fluidity and ion permeability [1] and, together with sphingolipids, in the organization of membrane domains or rafts [2]. These, in turn, regulate endocytosis [3] and act as intracellular signalling platforms for differentiation and survival pathways [2,4]. Intracellularly, cholesterol is a precursor of steroid hormones, vitamin D and oxysterols [5]. Moreover, the enzymes that produce the initial intermediates of cholesterol participate in the synthesis of vital classes of molecules, such as dolichol and ubiquinone [6] and, in addition, are used to modify small GTPases [7]. With such functional pleiotropism it is evident that cholesterol will be a major regulator of neuronal function, due to the constant demand of these cells for a tight control of ion balance, endocytosis and intracellular signalling following electrical stimulation. This especial need for cholesterol is evidenced by the fact that the brain contains 25% of the total pool of this lipid, even though it is only 2% of the total body weight [8]. Also, this special need makes obvious that any imbalance in brain cholesterol levels will have severe consequences to brain performance. The decay on such performance is one of the hallmarks of aging and, when exacerbated, of age-related neurodegenerative disorders. Here, we will review the data available on the influence that cholesterol alterations might have in normal and pathological aging with focus in Alzheimer's disease (AD).

Homeostasis of brain cholesterol: independence of the periphery

Outside the brain, cholesterol requirement is covered by de novo synthesis and by cellular uptake of dietary cholesterol under the form of lipoprotein-cholesterol complexes. In the brain, however, the bloodbrain barrier prevents the uptake of this lipid from the circulation. Several evidences demonstrate that the homeostasis of brain cholesterol is independent from the circulating cholesterol [8][9][10]. Thus, practically all (N95%) the cholesterol present in this organ is provided by de novo synthesis. This has been consistently shown in studies carried out in a variety of different experimental animals and also in humans [11][12][13]. Next we analyze data available on how brain cells deal with the synthesis and removal of cholesterol.

Synthesis of brain cholesterol

Since the peripheral and central cholesterol pools are not readily interchangeable, the brain has to produce cholesterol by de novo synthesis. This is an energetically expensive and complex pathway that requires several enzymatic reactions distributed in different organelles (see Fig. 1). The major rate-limiting enzyme in the cholesterol synthesis pathway is the endoplasmic reticulum-bound 3-hydroxy-3-methylglutaryl-coenzyme-A reductase (HMGCoA reductase) [9]. This enzyme is regulated by a potent negative feedback system that affects activity, stability and gene expression. Controlled degradation of the HMGCoA reductase [14], modulation of its mRNA translation efficiency (for a review see [15]) and regulation of gene transcription [16] have been reported.

From the HMGCoA reductase step cholesterol biosynthesis follows two alternative routes: one produces desmosterol as immediate precursor of cholesterol, the other gives 7-dehydrocholesterol as the direct precursor. Desmosterol is reduced to cholesterol by the enzyme 3β-hydroxysterol-Δ24 reductase (DHCR24/seladin-1). Mutations in the dhcr24 gene resulting in the limited ability to convert desmosterol to cholesterol, lead to a human metabolic disorder, desmosterolosis [17]. 7-dehydrocholesterol is converted to cholesterol by the 7-dehydrocholesterol Δ7-reductase. Mutations in the gene encoding this enzyme cause the human genetic disease Smith-Lemli-Opitz syndrome.

Approximately 70% of the brain cholesterol is associated to the myelin sheath, which consist of sections of oligodendrocyte plasma membrane wrapped around the axon. Studies performed in cultured cells have demonstrated that oligodendrocytes have the highest capacity for cholesterol synthesis, followed by astrocytes that synthesize at least 2-to 3-more cholesterol than neurons [18]. However, also neurons have a particularly high demand for cholesterol to form and maintain axons [19,20], dendrites [21,22] and synaptic connections [23,24]. It has been proposed that neurons rely the energetically expensive process of cholesterol synthesis on astrocytes [25]. Evidence for a cholesterol shuttle from astrocytes to neurons has been presented [23] and results obtained from in vitro systems have shown that neurons require cholesterol delivered by glial cells to form numerous and efficient synapses [25]. Later, by comparing cholesterol synthesis in cultures of immunoisolated cells from postnatal rats, Nieweg et al. [26] found that neurons and glial cells present distinct profiles of biosynthetic enzymes, post-squalene precursors and cholesterol metabolites. They have also demonstrated that neurons produce cholesterol inefficiently and show a much lower capacity to up-regulate the synthesis pathway than astrocytes, further contributing to the theory that neurons delegate this costly metabolic pathway to astrocytes.

Inside the brain, under normal conditions, a very efficient apolipoprotein-dependent process recycles cholesterol. The most abundant apolipoprotein is apolipoprotein E (ApoE) found in high-density lipoprotein particles. The differential expression of this protein in glia or neurons also evidences the distinct way by which brain cells manage cholesterol metabolism. Hence, while astrocytes possess the highest capacity for ApoE production [27,28], neurons can be induced to express ApoE under certain physiological and pathological conditions [29][30][31].

Cholesterol removal from the brain

Cholesterol in the adult brain is considered metabolically inert, however, a small fraction of the pool (≈0.02% in humans and ≈0.4% in mouse) turns over each day [32]. On the other hand, cholesterol synthesis in the developing CNS is relatively high, but it declines to a very low level in the adult [33]. To maintain the steady state, a small amount of cholesterol has to be continuously exported from the brain into the circulation. At the moment, two different mechanisms for cholesterol elimination from the brain are known. The major mechanism by which cholesterol is metabolized in the brain is by conversion to 24(S)-hydroxycholesterol [34] that diffuses out of cells, crosses the blood-brain barrier, and is cleared by the liver [35][36][37][38][39]. This reaction is catalyzed by a cytochrome P450 enzyme, Cholesterol 24-hydroxylase (CYP46), which is selectively expressed in the brain [40]. High levels of this enzyme are found in neuronal cells, specifically in pyramidal neurons of the hippocampus and cortex, in Purkinje cells of the cerebellum, and in hippocampal and cerebellar interneurons [40,41]. Interestingly, Bogdanovic et al. [42] have also reported CYP46 immunoreactivity in glial cells from brain of AD patients.

The other mechanism for cholesterol elimination, called reverse cholesterol transport pathway, involves the cholesterol transport across the plasma membrane to apolipoprotein acceptors in peripheral tissues [43,44] by ATP binding cassette (ABC) transporters. Members of the ABCA sub-family are expressed in isolated human neurons and in neuronal cell lines [45,46] and members of the ABCG sub-family have been described in brain and neuronal tissue [47,48]. Evidence proposing a role for ABCA1 and ABCG1 in the cholesterol efflux from astrocytes and microglia has also been presented [49,50].

All the above supports the current concept that the synthesis and elimination of cholesterol in the adult brain is compartmentalised. Astrocytes are believed to be responsible for the majority of synthesis in the adult brain while they contribute relatively little to brain elimination. On the other hand, the neuron specific cholesterol 24hydroxylase, CYP46A1, is responsible for the elimination of about two thirds of the cholesterol synthesised by the brain [37,51]. Indeed, knockout (KO) mice lacking CYP46 have a ≈50% reduction in brain cholesterol excretion [52]. This decrease is compensated by the reduction in de novo synthesis, resulting in steady state levels of cholesterol in the brains of KO mice that are similar to those of wildtype (WT) mice. This reduced synthesis is likely to be mediated by a decrease in the activity of HMG CoA reductase. Interestingly, using learning tests in mice and electrophysiological experiments with hippocampal slices in vitro, Kotti et al. [53] showed that cholesterol turnover via the CYP46 ensures activation of the mevalonate pathway and the constant synthesis of geranylgeraniol, which in turn is essential for learning and memory.

Changes in brain cholesterol homeostasis during normal aging: causes and consequences

Aging is characterized by a decline in cognitive functions such as reduced learning ability and memory [54,55]. In early reports these deficits were linked to neuronal loss in the hippocampus, one of the brain regions where task memory centres are located [56][57][58][59]. These observations were consistent with subsequent data suggesting that neuronal death during senescence could be due to the decrease in the availability of growth factors [60][61][62][63]. Conversely, a plethora of recent evidences show that neuronal density in the hippocampus is unaffected during aging in both, animal models and humans, despite the decrease in neurotrophin concentration [56]. This last scenario indicates that post-differentiated neurons must possess a robust and long-term survival strategy to outlive the noxious effects of trophic factor deficiency under the prolonged pressure of multiple stress insults, which act as major determinants of aging [64][65][66][67]. The mechanisms behind survival under such adverse conditions remain largely undefined. Nonetheless, it has been shown that during aging hippocampal neurons, in vitro and in vivo, activate the TrkB/Akt survival pathway in a ligand-independent manner, by virtue of the loss of membrane cholesterol levels [4]. In agreement, a moderate loss of brain cholesterol with age has been described [59]. However, other evidences support that there is not an age-dependent change in the absolute amount of cholesterol in the hippocampus [68] or that cholesterol content of the exofacial leaflet of synaptic plasma membranes doubles with age [69,70]. It is presumable that cholesterol metabolism in the different regions of the brain is not uniform [71] and that during normal aging brain cholesterol levels would differentially change depending on the region considered. Indeed, differences in cholesterol content have been found depending on the brain region and cell-type studied [72], and also differences in lipoprotein transporter and receptors distribution [73], as well as in the expression of cholesterol synthesizing enzymes [74], have been described. It was also shown that cholesterol synthesis is decreased in the human hippocampus, while absolute cholesterol content remains stable [75]. On the other hand, a direct correlation between cholesterol synthesis rate and hippocampal activity has been proposed. According to that, impairment of cholesterol synthesis or lipoprotein transport diminishes synaptic plasticity and therefore cognitive functions [53,76,77].

All the presented data indicate that changes in cholesterol homeostasis occur during aging in different brain areas and that these changes may influence the survival capacity of neurons or their performance. Among the cholesterol related molecules, two have been especially linked with these aspects: the transporter molecule ApoE and the cholesterol-removing enzyme CYP46.

ApoE and aging

The observations that ApoE4 is a risk to suffer AD and that this allele is not only less concentrated [78] but also shows low cholesterol transport capacity [79], together with data describing that aged ApoEnull mice are more susceptible to neurodegeneration [80,81], led to think that ApoE deficiency could contribute to the aging process. ApoE-null mice have been analyzed in this context. Some studies in ApoE-null mice did not find signs of synaptic degeneration [82][83][84][85][86][87] but others reported that ApoE-null mice develop severe learning and memory deficits [88], probably associated to cholinergic deficiency [89,90] and to the presence of synaptic changes [91][92][93][94].

Studies showing that ApoE expression increases in the liver in an age-dependent manner [95] suggested that ApoE levels might also change in the aged brain. This, however, is still controversial. Hence, studies in mice showed a marked decrease of ApoE expression in the aged hypothalamus and cortex [96], but increase in the hippocampus [97]. In rats, expression of ApoE in glial cells is elevated during aging in a region-specific manner [98,99]. However Gee et al. found no change in ApoE mRNA and protein levels either in the cortex, striatum or hippocampus of aged rats [100]. On the other hand, Kadish et al. [101] have described a midlife upregulation of ApoE and proposed that a new source of excess free cholesterol develops, possibly as a byproduct of lysosomal degradation of cholesterol-rich myelin fragments, in turn activating upregulation of ApoE and other cholesterol trafficking genes/proteins in astrocytes. They postulate that transport of cholesteryl esters in ApoE-containing lipoproteins from astrocytes to remyelinating oligodendrocytes may mediate incorporation and long-term storage of excess cholesterol in new myelin.

CYP46 and brain aging

The role proposed for CYP46 in brain aging is based on the changes observed in the levels of this cholesterol-removing enzyme in the aged brain and on its consequences. Hence, Lund et al., [40] showed an increased amount of CYP46 in brains of old humans and mice compared to young ones. Evidences from our laboratory have demonstrated that this age dependent increase in CYP46 levels results in a decrease of membrane cholesterol in hippocampal neurons in vitro and in vivo [4]. This loss of cholesterol triggers the activation of the TrkB/Akt prosurvival pathway, protecting neurons in conditions of high stress in vivo and in vitro [102]. This appears consistent with data from CYP46KO mice, Halford et al. [103] demonstrating that reduction of cholesterol synthesis rate in the mouse brain extends lifespan. Although total cholesterol levels in the brains of these KO mice do not change [51,52], the variations on cholesterol levels in the membranes of specific neuronal populations or in old CYP46KO mice are unclear.

Understanding the causes for the age dependent variation of CYP46 levels is still a challenge. Relevant in this regard are the studies of the mechanisms controlling its expression. Ohyama et al. [104] found that the expression of cyp46A1, the gene encoding CYP46, was insensitive to sterol levels, the normal mechanism regulating HMGCoA reductase and CYP7A1, two key genes in cholesterol synthesis and elimination, respectively. On the other hand, data exist indicating that cyp46A1 can be induced under certain stress conditions such as cortical injures, induced autoimmune encephalomyelitis or in AD [42,105,106]. A recent study by Milagre et al. indicated that members of the SP1 family of transcription factors might regulate cyp46A1 [107]. In other work performed by Shafaati et al. [108], a marked time-dependent derepression of the expression of cyp46A1, in response to treatment with the histone deacetylase (HDAC) inhibitor Trichostatin A was demonstrated. Especially relevant in the context of aging is the finding that oxidative stress is a potent inductor of CYP46 expression [104].

Consequences of CYP46 upregulation

It is important to remark that 24S-hydroxycholestrol and other oxysterols are repressors of genes involved in cholesterol synthesis or uptake [109]. They also act as activators of liver-X receptors (LXRα and LXRβ) [110,111], which in turn regulate different genes involved in lipid homeostasis [112,113]. For example, activation of LXR induces the expression of ATP-binding cassette transporters [114,115] and treatment with LXR agonists induces the expression of ABCA1 and ABCG1 and results in increased cholestrol efflux in astrocytes. A much lower effect was observed in primary neuronal cultures [114,116] and according to that, ABCA2 transporter, predominantly expressed in neurons, seems not to modulate cholesterol efflux [44].

Clement et al. [117] have shown that oxidative stress induces accumulation of 24S-hydroxycholesterol and down regulation of LDL receptor-related protein (LRP, one of the main ApoE receptors in neurons), HMGCoA reductase and ABCA1 in murine hippocampal nerve cells. In agreement to these results, in neuroblastoma cells, 24Shydroxycholesterol accumulation decreased mRNA levels of the cholesterol synthesis genes HMG CoA reductase, squalene synthase and farnesyl diphosphate synthase but did not alter levels of the mRNA of fatty acid synthesis genes. According to those results, traumatic brain injury induced CYP46 and resulted in a significant decrease of HMG CoA reductase and squalene synthase mRNA level [118].

Thus, it appears that induction of CYP46 during aging would lead not only to increased cholesterol hydroxylation and reduced uptake by neurons but also to reduced synthesis in astrocytes and increased efflux of cholesterol by reverse cholesterol transport through ABCA1 transporter.

Alterations in brain cholesterol homeostasis in pathological aging: implications for AD

In the previous sections we have reviewed data on the appearance, causes and consequences of brain cholesterol changes in the normal aging. However, cholesterol alterations have been also reported in several neurodegenerative disorders associated with age [119]. A strong link between this lipid alterations and pathology has been made for AD. In fact, the only two established risk factors for the late onset forms of this disease, which represent the vast majority of the cases, are aging and the inheritance of the e4 allele of the cholesterol transport protein ApoE [120]. This last finding fostered the research on the link between cholesterol and AD for the last 15 years. Today, many other cholesterol-related molecules have been associated to the disease. These participate in practically all aspects of cholesterol metabolism including synthesis, uptake, distribution or removal. Next, we summarize and discuss the evidences supporting such associations as well as their possible relevance for non-pathological aging.

AD and cholesterol synthesis

The first link between cholesterol synthesis and AD came from the finding that hypercholesterolemic patients treated with statins showed a drastic reduction in AD prevalence [121,122]. Statins are inhibitors of the HMG-CoA reductase, the enzyme that catalyzes the rate-limiting step of cholesterol synthesis [123]. A first hypothesis derived from these results, supported by data in vitro and in vivo [123][124][125], proposed that high brain cholesterol would contribute to increase the amyloid peptide (Aβ) production and thus disease [126]. Statins would exert their beneficial role by lowering cholesterol levels in the brain. This assumption is nowadays difficult to maintain because these drugs are most efficient in reducing plasma cholesterol levels but are unlikely to affect the brain levels of this lipid [127]. Currently, the most accepted view is that statins avoid brain functional decay with aging and are protective against AD because, by reducing peripheral cholesterol, they favour brain oxygenation and therefore a better clearance of the amyloid peptide to the blood stream. Statins might also delay brain aging and pathology because of their anti-inflammatory action [127].

A differential display to identify genes with altered expression in the brain regions most vulnerable to AD provided another link between AD and cholesterol synthesis. This identified a marked reduction of the gene encoding for the 3β-hydroxysterol Delta 24reductase in the affected areas [128]. This gene codifies for the enzyme that catalyzes the last step of the desmosterol to cholesterol pathway [17], which has also received the name of seladin1 (for SELective AD indicator 1). The finding that a cholesterol-synthesizing enzyme was reduced where AD pathological signs are most evident led to propose the opposite hypothesis to the one mentioned above: that brain cholesterol reduction, and not increase, would be a trigger for AD. Although it is still not clear whether reduced Seladin1 levels are a cause or consequence of the disease several evidences support a causal role for this protein. On one hand, seladin1 controls cholesterol levels at the neuronal membrane thus contributing to the maintenance of its compartmentalization. The presence of APP and its amyloidogenic secretases appear to be compartmentalized. While the β and γ secretases seem to be predominantly localized in cholesterol enriched microdomains [129,130], there is controversy about APP distribution. Hence, while some reports show its localization in cholesterol enriched domains [129] others locate this protein outside them [131]. The use of different cell types may explain this apparent contradiction. In cultured neurons and in mice brains it has been shown however that reduction of cholesterol levels, induced pharmacologically or due to Seladin1 deficiency, impaired the membrane segregation of APP with its β-secretase BACE, leading to their increased interaction and enhanced amyloid production [131,132]. Compartmentalization of the membrane seems to be also required to activate the amyloid-degrading enzymes plasmin and insulin degrading factor [131,133,134], whose activities are reduced upon seladin1 silencing [132,134]. Hence, seladin1 deficiency could contribute to amyloid accumulation by increasing the production and impairing the degradation of the peptide. It is not known whether seladin1 levels are also reduced in non-disease aged brains compared to young ones. If this were the case, it could be proposed that membrane compartmentalization loosening accompanies brain aging and that this would have deleterious effects on many signalling processes, which largely rely on membrane domains, thus contributing to functional decay (see Fig. 2). On the other hand, it is important to note that seladin 1 has an antiapoptotic action. By inhibiting caspase 3, seladin 1 confers protection against Aβ induced apoptosis and oxidative stress [135]. Thus, low levels of this enzyme would enhance the vulnerability of cell populations to stress accumulated during aging and/or caused by AD related insults.

AD and cholesterol transport and uptake

The involvement of alterations in cholesterol synthesis in AD is clearly worth it to study, however, it is important to keep in mind that, as neurons age, de novo synthesis of cholesterol is progressively switched off and glial cells become the main source of this lipid [136]. Therefore, aging renders neurons dependent on the transport and uptake of cholesterol from glia. ApoE and its receptors, the lowdensity lipoprotein receptors (LDLR), respectively, mediate these events. Associations of both types of molecules with AD have been found. The first evidences are genetic. Hence, carriers of the e4 allele of the ApoE protein have increased risk to suffer the disease and an earlier onset than those individuals with e3 and e2 alleles [120,137]. On the other hand, polymorphisms within LDLR genes (i.e. LRP1 and apoER2) have been also associated with a higher predisposition to suffer AD [138,139] (a list of polymorphisms associated to the disease in these and other cholesterol related molecules is provided in Table 1). Studies in vitro and in vivo have tried to explain the reasons for such risk/predisposition. These are reviewed next.

The e4 allele of ApoE may prompt to the disease because of deficient metabolism of cholesterol. A series of findings support this view. ApoE4 individuals [140] and mice [78] show low concentrations of the protein in plasma and brain, respectively, compared to the other genotypes e2 and e3. Moreover, the e4 genetic variant has a poor response to physiological inducers of expression [141] and a lower ability to release cholesterol from glial cells [79]. This could result in lower levels of cholesterol in neuronal membranes that could, in turn, have consequences in the compartmentalization of APP and its secretases and/or in the activation of amyloid degrading enzymes, similar to what it has been mentioned above for seladin1 deficiency. Besides its influence on cholesterol levels ApoE has other roles that, when altered, could contribute to disease. Hence, ApoE has antioxidant properties [142], binds to Aβ favouring the removal of the peptide from brain into the general circulation [143] and have been related to synapse maintenance and cognition [144]. Decreased levels or deficient activity of this protein, which seems to characterize the e4 allele, would have deleterious consequences in any of these events. Finally, atherosclerosis commonly affects apoE4 bearers. The formation of cholesterol deposits in brain arterioles and capillaries leads to hypo-oxygenation, which could impair brain function and clearance of Aβ [145].

The LDLR family mediates the entrance of cholesterol into neurons by binding to ApoE [146]. Hence, alterations in these molecules affect cholesterol levels in these cells that would, in turn, contribute to pathology. In addition to promote changes in cholesterol levels alterations in LDLR could contribute to AD pathology by other means. Thus, several members of the LDLR family (i.e. LRP1B, SorLA/LR11 or ApoER2) interact with APP and regulate its rate of internalization [146][147][148], which is a critical step in the amyloidogenic processing. LDLR can also interact with components of the secretases involved in APP cleavage, influencing APP access to them [149,150]. At least one of the LDLR family members, LRP1, mediates clearance of the amyloid peptide in vitro [151] and is a major efflux transporter of the peptide across the brain blood barrier [152], therefore favouring its removal from the brain. Importantly, LDLR members mediate signalling events necessary for synaptic plasticity and survival, which involve a number of kinases [153]. It has been shown that alterations in the levels of LDLR molecules, such as ApoER2, lead to the hyperphosphorylation of Tau protein [154], which together with amyloid accumulation is a hallmark in AD.

AD and cholesterol intracellular distribution

Recent evidence on the effects that the modulation of the Acylcoenzyme A cholesterol acyltransferase (ACAT) has on amyloid accumulation highlighted that it is not only the levels of cholesterol what might be key in AD pathology but how this lipid is distributed inside the cells. ACAT controls the relative amounts of free cholesterol and cholesterol esters by catalyzing the formation of the latter from cholesterol and fatty acids in the endoplasmic reticulum [155]. While free cholesterol distributes mainly to the plasma membrane, endosomes and Golgi apparatus, cholesterol esters are stored in cytoplasmic lipid droplets [156]. It has been shown that silencing of ACAT, through genetic or pharmacological modulation, results in a drastic reduction of Aβ production in cell lines [157]. Inhibitors of ACAT activity have been also tested in vivo being able to reduce cholesterol ester levels and soluble and aggregated amyloid peptide in the brains of AD mouse models [158]. Remarkably, spatial learning was improved in these animals [158]. It seems then logical to propose that inhibition of this enzyme could ameliorate not only the cognitive deficits in AD patients but also age associated brain functional decay. How intracellular cholesterol distribution compares between aged and young neurons or whether the brain levels of ACAT change upon non-pathological aging has not been determined. Interestingly, analysis performed in rodent liver did reveal an increase of the enzyme with age [159].

AD and cholesterol removal

As mentioned in earlier sections there are no means to degrade brain cholesterol. This is why its transport across the plasma membrane and/or the transformation of the molecule are the ways by which brain cells remove used cholesterol. ABCA1, which has a channel-like structure that can transport solutes across the cell membrane, mediates the transfer of cellular cholesterol onto lipidpoor apoE [160] and CYP46 hydroxylates the lipid converting it in a soluble metabolite, 24S-hydroxycholesterol, which can freely cross the brain blood barrier [40]. A correlation in certain families between polymorphisms in both ABCA1 and CYP46 and the risk to suffer lateonset AD has been reported [161,162]. In the case of CYP46 polymorphisms the risk synergistically increases with the additional presence of 1 or 2 apoE4 alleles [162]. Moreover, it was shown that they are associated with increased Aβ load in brain tissues as well as with increased cerebrospinal fluid levels of the peptide and phosphorylated tau protein [162].

Changes in the expression of ABCA1 levels affect both central and peripheral cholesterol. On one hand, this cholesterol transporter is highly expressed in the CNS [45] and mice lacking it show reduced apoE in the brain and cerebrospinal fluid and low levels of cholesterol [163]. On the other hand, mice lacking ABCA1 specifically in the CNS presented reduced plasma high-density lipoprotein (HDL) cholesterol and enhanced brain uptake of esterified cholesterol from plasma HDL [164]. Changes on ABCA1 expression also modulate the amount of Aβ peptide most probably in a cholesterol dependent manner. Hence, induction of its expression by activation of liver X receptors (LXR) or by transfection inhibited Aβ production in vitro and in vivo [44,165].

Consistently, genetic loss of LXR receptors or ABCA1 increases Aβ load [166,167].

The specific presence of CYP46 around neuritic plaques in AD patient brains is one of the non-genetic evidences supporting the involvement of this enzyme in the disease [168]. Moreover, mice in which CYP46 is missing exhibit deficiency in learning and memory [53]. The fact that, as already mentioned, this cholesterol hydroxylase mediates response to stress by promoting TrkB survival pathways [102] also suggests that its deficiency would make neurons more susceptible to stress in normal or pathological aging. This conclusion would be consistent with data showing that oxidative stress is a potent inductor of CYP46 expression [104]. It is worth mentioning that the product of CYP46 activity: 24-hydroxycholesterol, has also been related to AD. The amount of this oxysterol is diminished in AD brains [169] and it inhibits Aβ production at least in vitro [168]. It has been proposed that this oxidized cholesterol favours the nonamyloidogenic processing of APP [170] although the mechanism of action is not yet clear.

Cholesterol homeostasis: target to fight brain aging and associated disease

From all the above it comes clear the message that cholesterol imbalance will have drastic consequences. This has moved many to develop strategies to modulate cholesterol regulatory molecules to prevent, diagnose, slow down or even cure AD. Because many regard this disease as an exacerbation of normal aging, and because changes in cholesterol occur in the latter as well, modulation of this lipid levels is also envisioned to improve the performance decay occurring in a healthy brain with age. The comparison of cholesterol levels in young vs aged brains and neurons [4] and between aged-matched control vs AD patient brains [133] showed the existence of a moderate loss of this lipid (around 20%) during normal aging, exacerbated in disease. While cholesterol loss in the aged has been shown to improve survival [4,102], a reduction of this lipid, beyond 30%, can lead to cell death. It is important to keep this in mind when deciding to put somebody under statin treatment, as these may lead to a pathological cholesterol loss. Having said this we will next review the strategies that have been or are being considered to modify cholesterol levels based in the modulation of the cholesterol metabolic events and related molecules that so far have been related to AD.

Modulating cholesterol synthesis

Statins, the inhibitors of the cholesterol-synthesizing enzyme HMG-CoA reductase, are widely used for the treatment of hypercholesterolemia in humans. Since their ability to reduce the incidence of AD was discovered, numerous studies have been performed aiming to validate their therapeutical value against AD. The conclusions from prospective cohort studies and clinical trials are however not as clear as that aroused from the retrospective studies performed in hypercholesterolemic patients. Hence, treatments with different statins for different periods of time showed no significant or just temporary cognitive improvement, which in certain cases were not dependent on lipid levels [171][172][173]. Although these drugs might not serve to fight AD specifically it is likely that their anti-inflammatory actions and their capacity to reduce peripheral cholesterol, thus contributing to a better oxygenation of the brain, will be beneficial to maintain brain function with time.

Preventing the downregulation of the cholesterol-synthesizing enzyme, seladin1, appears to be a good strategy to diminish stressinduced apoptosis and the loosening of membrane compartmentalization and thus amyloid accumulation and impaired signalling. Recent studies have shown that estrogens up regulate seladin1 expression, possibly by activation of the functional estrogen responsive elements upstream of the coding region of its gene [174,175]. ⁎ Note that for all the studied genes the linkage of a given polymorphism to AD depends on the population analyzed. Different results were obtained depending not only on the ethnic group (Caucasian, African, Asian, Hispanic and Mixed) but also on the nationality in the same ethnic group (for detailed information see [187]).

Despite the lack of consensus several studies have indicated that estrogen treatment may actually decrease the risk or delay the onset of AD in post-menopausal women [176] and that an earlier and more prolonged administration further reduce the risk [177]. Importantly, hormone therapy could also be considered a way to delay functional decay with age especially in women. In support, different studies have shown that estrogens improve memory, executive function and attention processes after menopause [178,179]

Modulating cholesterol distribution and uptake

The positive effects of ACAT inhibition in reducing amyloid load in cell culture and mice brains together with the learning improvement reported in mice models for AD treated with the ACAT inhibitor CP-113, 818 [158], strengthen the notion that modifying the intracellular distribution of cholesterol, through modulation of ACAT, might be a good strategy for the treatment and prevention of AD. Although results on clinical trials with inhibitors of this enzyme are not yet available, the efficient and non adreno-toxic ACAT inhibitor Avasimibe is considered safe for human use [179] and opens interesting perspectives to preserve cognitive capacities during aging and in AD.

A clear target to modulate cholesterol transport and uptake is ApoE, the current idea being to increase its levels and/or improve the function. With this aim large screenings of compounds have been performed, resulting in the identification of several enhancers of apoE synthesis and secretion [15]. From them, estrogen, the cyclooxygenase inhibitor indomethacin, the cholesterol lowering drug probucol and the peroxisome proliferating activating receptor-g (PPAR-g) agonist rosiglitazione have been utilized in placebo controlled, double blind clinical trials in mild-to-moderate AD patients. These compounds promoted the stabilization of the cognitive symptoms or even cognitive improvement [15,180,181]. Another strategy to favour the uptake of cholesterol in brain cells is the pharmacological modulation of the transcription factors that regulate the expression of the ApoE receptors of the LDLR family. Moreover, the use of the recombinant receptor-associated protein (RAP), which antagonizes ligand binding to LDLR family assisting their proper folding and trafficking along the secretory pathway [182], has been explored. Long-term treatment with RAP disrupts the intracellular interaction between LRP1 and APP, increasing cell surface APP and decreasing Aβ production [183].

Modulating cholesterol removal

As already mentioned CYP46 is responsible for most of cholesterol removal from brain [184]. Moreover, CYP46 is responsible for the cholesterol loss of aged neurons, in turn triggering the activation of the survival pathways mediated by TrkB [102]. This together with the observation that polymorphisms in its gene have been associated with AD [162], suggest that modulation of CYP46 would have influence in normal and pathological aging. Histone deacetylase inhibitors (e.g. valproate, vorinostat) are the only class of compounds known to enhance mRNA expression of the enzyme [108]. Treatment of AD mouse models with the HDACi valproate reduced Aβ production, neuritic plaque formation and behavioural deficits [185]. This would be consistent with data showing that enhancement of CYP46 plays an anti-stress role. On the other hand, too much activation of this enzyme poses the risk of an excessive loss of cholesterol that would have deleterious consequences in neuronal membrane function. Thus, a tight control of CYP46 activity would be probably essential to maintain the balance between healthy and pathological brain aging.

Cholesterol metabolism as a diagnostic tool

A major drawback to prevent or cure AD is the fact that diagnosis is only possible at very advanced stages of the disease when brain damage is probably irreversible. Thus, major efforts have been done to find early markers for the disease. Cholesterol metabolism seems to help in this matter. Hence, 24S-hydroxycholesterol levels in the cerebrospinal fluid have about the same diagnostic sensitivity as the standard biomarkers currently used in the diagnosis of neurodegenerative diseases and dementia namely, levels of tau protein, phosphorylated tau and Aβ [186]. Although elevated levels may simply reflect high stress, rather than being a real AD diagnostic marker, it is certain that they can be used for prognosis purposes to evaluate the evolution of any given treatment. For the same reason this cholesterol metabolite could serve to monitor the course of nonpathological brain aging as well. In fact, changes in the levels of 24Shydroxycholesterol in plasma and cerebrospinal fluid have been reported with age [38].

Closing remarks

Ever increasing evidences support the influence of cholesterol changes in the process of normal aging and in the ethiology of agerelated neurodegenerative diseases such as AD. Still, many open questions remain. Are altered brain cholesterol levels cause or consequence of aging and/or AD? Are brain cholesterol changes in the brain of aged individuals or AD patients a primary defect or an anti-stress response due to hypoxia from hypercholesterolemia or other systemic alterations? Which is the threshold in these changes that determines normal or pathological conditions? How are the changes in brain cholesterol leading to disease, by increasing amyloid production, reducing its clearance or altering ion permeability or signalling pathways? Are the main consequences in plasma membrane-controlled activities or is the intracellular cholesterol the one that matters? Answers to these questions, among many others, are an essential step before we could safely try cholesterol regulatory drugs to delay aging or treat AD in humans. Because of the many roles of cholesterol it is not unlikely that a cholesterol synthesis inhibitory drug showing a "good" effect, for example inducing a reduction in amyloid production, has the "bad" effect of impairing electrical transmission, because of altering ion permeability. Further research will contribute to clarify the intriguing role of brain cholesterol in normal and pathological aging.